On 25 March 2013 17:45, Martin J. Evans <boh...@ntlworld.com> wrote: > On 25/03/13 14:53, demerphq wrote: >> >> On 25 March 2013 15:48, Martin J. Evans <boh...@ntlworld.com> wrote: >>> >>> as we are producing massive JSON files and every extra "" makes a big >>> difference. >> >> >> If there is no specific reason to use JSON (like for instance sending >> data to a web browser) and size and speed are priorities then you >> might want to consider using Sereal instead of JSON. It is both faster >> and produces smaller output than JSON. >> >> Yves > > > Thanks Yves. I've tried Sereal but unfortunately it is just too late to > change now. Other reasons are a) it is not readable b) last time I looked it > was not supported by all the languages we need to support c) it is still > marked experimental.
Re a-c... A) its not readable, and it wont ever be. OTOH, would we be able to mitigate this factor for you if there was an easy to use to tool which would do something like translate a Sereal dump on the fly to something readable? You arent the first who has said that, and I can sympathize (although i have gotten surprisingly good at decoding sereal by sight :-), so i am wondering if there is a middle ground on this one that might tip the balance... (assuming the other points werent an issue). B) our cross language support is growing but still pretty small. By that standard I can understand sticking to JSON. OTOH, id be curious to know which languages are important to you. C) the experimental marking these days is mostly there so people give us feedback when they use it. We should probably change that advisory. (TBH, until you mentioned it I had forgotten it entirely.) Anyway, thanks for the feedback! Yves -- perl -Mre=debug -e "/just|another|perl|hacker/"
